I'm trying to prove the inequality

$$\int_{0}^{1}xf(x)dx \leq \frac{2}{3}\int_{0}^{1}f(x)dx$$

for all continuous concave functions $f: [0,1]\rightarrow [0,\infty)$. I've been working on this for a while and would just love a hint if anyone can see something obvious that I'm missing.

What I've tried is observing that $\frac{2}{3} = \int_{0}^{1}\sqrt{x}dx$ and trying to use Holder's inequality with $p=q=2$ on the integral $\int_{0}^{1}xf(x)$ but the squaring and the integral seem to be in the wrong order. I think there might be something more to this idea since that's the only way I know off the top of my head that can split an integral of a product into a product of the integral at the cost of inequality. However, I can't see how concavity comes into play.

Something else I've atempted is observing that because of the concavity of $f(x)$ we have for all $x \in [0,1]$ that $$xf(x) + (1-x)f(0) \leq f(x^2 + 0 (1-x)) = f(x^2)$$

so that $$\int_{0}^{1}xf(x) + (1-x)f(0) \leq \int_{0}^{1}f(x^2) \leq f\left(\int_{0}^{1}x^2dx\right) = f(1/3)$$ where the second to last inequality is Jenson's inequality (since $f(x)$ is concave). Evaluating the first integral as much as possible gave $$\int_{0}^{1}xf(x)dx + f(0)/2 \leq f(1/3)$$ but the above result feels incorrect somehow eventhough I feel fairly confident in every step.


We just want to show $$ \int_0^1 (x-2/3)f(x)dx \leq 0$$ We have \begin{align} \int_0^1 (x-2/3)f(x)dx &= \int_0^{2/3} \underbrace{(x-2/3)}_{\leq 0}f(x)dx + \int_{2/3}^1 \underbrace{(x-2/3)}_{\geq 0}f(x)dx \\ &\leq \int_0^{2/3}(x-2/3)L(x)dx + \int_{2/3}^1(x-2/3)U(x)dx \end{align} where $L(x)$ and $U(x)$ are any lower-bound and upper-bound functions over their respective intervals, so that \begin{align} L(x) &\leq f(x) \quad \forall x \in [0,2/3] \quad \mbox{(Eq. 1)}\\ U(x) &\geq f(x) \quad \forall x \in [2/3,1] \quad \mbox{(Eq. 2)} \end{align} But by concavity and nonnegativity of $f$, the linear function $$L(x) = U(x) = x\frac{f(2/3)}{2/3} \quad \forall x \in \mathbb{R}$$ satisfies (Eq. 1) and (Eq. 2). (Just draw a picture and note that a concave function lies above its chords.) So $$ \int_0^1 (x-2/3)f(x)dx \leq \int_0^1(x-2/3)x\frac{f(2/3)}{2/3}dx = 0$$ The bound is tight because it holds for the linear function $f(x)=x$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.